Optimizing Training and Feedback for Spatial Intercell Interference Cancellation

In this paper, we investigate spatial intercell interference cancellation - an efficient technique to mitigate intercell interference in multicell networks. We consider a practical model for channel state information (CSI), where the transmit CSI is acquired through downlink training and uplink feedback. Due to the requirement of channel information from multiple base stations, the training and feedback design is quite different from conventional single-cell processing systems. We optimize training and feedback, where both analog and digital feedback is considered. For analog feedback, it is shown that the downlink training optimization provides a more significant performance gain than feedback optimization; while conversely for digital feedback over a finite-rate feedback channel, the feedback bit allocation is more important than the training optimization.